Measurement in Financial Reporting

Abstract

Much of the current thinking of accounting standard setters about measurement seems to be based on an idealized view of markets as being complete and in perfectly competitive equilibrium. In such conditions, there is a unique market price based on full information for every asset and liability, and there is an obvious attraction in using this price as the measure for accounting. This explains the apparent enthusiasm for single ideal measurement methods based on market price, such as fair value (FV), defined in SFAS 157 as a market selling price. In reality, markets are imperfect and incomplete, so that ideal unique market prices are not available for all assets and liabilities. This is why SFAS 157 (FASB, 2006) has to resort to a Fair Value Hierarchy, the lower levels of which require estimation of what a market price might be if one existed. The liquidity problems evident in the GFC have demonstrated the limitations of markets and the difficulty of estimating market prices, even for financial instruments. An important source of imperfection in markets is the existence of information asymmetry, which means that not all market participants are equally well informed. The alleviation of information asymmetry through the exercise of accountability is the principal reason why accounts are needed. Measurement methods should be selected with the market context in mind. Having a single preselected measurement method may not best reflect market conditions or meet users' needs. Accounting provides information as input to users' own decision models, rather than providing a decisive valuation. The current project of the IASB and the FASB to develop a joint conceptual Framework (CF) started by following the approach of the 2005 IASB Discussion Paper (IASB, 2005, prepared by staff of the Canadian Accounting Standards Board) in attempting to identify the single measurement basis that best conforms with criteria such as relevance and representational faithfulness that are prescribed by the CF. These measurement bases are ‘pure’ bases of measurement such as historical cost, replacement cost, value-in-use or fair value. The objective would always be to estimate the selected measurement basis, other bases being allowed only as proxies where direct measurement was impossible. An example of the use of proxies is the Fair Value Hierarchy adopted by the FASB in SFAS 157, where fair value (selling price) is the measurement basis, but estimates using other information are permitted when selling prices cannot be observed directly. This approach does not favour mixed measurement approaches, such as the traditional ‘historical cost or market value, whichever is the lower’ or the more recent deprival value method which provides an algorithm for selecting from a range of current values, because these mix different ‘pure’ measurement bases, albeit in a systematic way. This mixing of different measurement methods is believed to create mismatch problems, where different items in the same set of accounts are measured on a different basis, so that aggregation is misleading. However, as Mary Barth indicates, the two Boards' recent deliberations have considered a new approach which might allow mixed measurements. The single method approach would certainly promote consistency within accounts, avoiding mismatches and allowing more meaningful aggregation, although aggregation problems are also present within each ‘pure’ approach: The whole can rarely be measured as the sum of the parts. This approach would also improve comparability across entities, although comparability is of limited value if the data being compared are not relevant to the user. In the case of fair value, which has tended to be favoured as a pure measure, in standard setters' deliberations the claim has been made that this method does have the property of relevance because it measures the market's expectation of future cash flows (see Barth, 2006, 2007; Hague, 2007: the former author was an IASB Board member and the latter one of the team that developed the 2005 Discussion Paper). It is also claimed that it has the property of objectivity, reflecting the market's view rather than the entity-specific views of managers (IASB, 2005, Discussion Paper, Ch. 4). The adoption of a single measurement method is predicated on the belief that such a measure will be always the most relevant and will be reliably measurable. This will be the case if markets are complete and in perfectly competitive equilibrium. In such a situation, a unique market value can be attributed to every asset and liability, so a single measurement method, consistent with FV (but without the need to discuss the choice between entry and exit values, because the two are now identical under the ‘law of one price’) is appropriate. Such properties as consistency and comparability can then be achieved in a very precise sense. Unfortunately, in reality, markets are not perfect and complete, so that this ideal information is not available. Grossman and Stiglitz (1980) have demonstrated that information asymmetry, which is pervasive in markets and whose partial relief is a fundamental object of accounts, is a fundamental barrier to market perfection: ‘because information is costly, prices cannot perfectly reflect the information which is available, since, if it [sic] did, those who spent resources to obtain it would receive no compensation’ (p. 405). In a similar vein, Beaver and Demski (1979) have pointed out that, in a world of perfect and complete markets, accounting would not be required, because everybody would be fully informed, so that the very existence of accounting implies a degree of market imperfection. Beaver and Demski's response to this dilemma is to suggest that, in a realistic setting of market imperfection, we should regard accounts as providing useful information for decision-making rather than definitive measurements. In practical terms, this means abandoning the standard setters' search for the single ideal measure and adopting instead the objective of identifying the information that is most likely to serve the needs of users' decision models. In particular, the informational approach is not concerned to value the business, or a part of it, by measuring the present value of the future cash flows, but rather to provide information that will help users such as financial analysts to perform their own valuations, for example, by providing information that will assist the prediction of future cash flows. The nature of such information is likely to be more eclectic and more dependent on specific circumstances than that implied by the ‘single ideal’ approach. However, it does not preclude the universal use of a single measure if that is justified by users' information needs and by the practical constraints of accounting. For example, Edwards and Bell's (1961) replacement cost model can be regarded as adopting an informational approach (Whittington, 2008). The recent GFC has provided a vivid illustration of the practical importance of market imperfection and incompleteness. Financial markets were believed by standard setters to be commonly ‘deep and liquid’ and financial instruments were therefore selected, in IAS 39 (IASB, 2008), as being, in many cases, suitable for FV measurement. However, in the course of the crisis, the markets for many financial instruments have become very illiquid and, in some cases, non-existent, leading the FASB and the IASB to relax their FV reporting requirements. Fair value epitomizes the concept of an ideal single measurement objective which is believed to be intrinsically superior to others. This belief is predicated on the assumption of an idealized complete perfect market equilibrium, where market prices are readily observable for all assets and liabilities (markets are complete) and reflect the cash flows expected to be obtained by the fully informed marginal trader (perfectly competitive equilibrium). The market prices required by fair value are non-entity specific, that is, they do not reflect the specific circumstances and economic opportunities facing the entity holding the items being measured. Advocates of fair value regard this as a virtue, because entity-specific measurement may introduce into measurement the subjective estimates of management. However, the belief that such subjective estimates will have to be made does contradict the assumption that markets are perfect, and therefore fully informed. In reality, accounts are concerned with reporting the state of a specific entity, in order to fulfil the accountability function to users who are not fully informed. In this context, reporting the opportunities available to the specific entity seems to be desirable and probably more useful than reporting the hypothetical opportunities available in a theoretically constructed market (as in the lower levels of the Fair Value Hierarchy). Fair value as defined in SFAS 157 has, in consequence of its underlying perfect market assumptions, a number of deficiencies from the perspective of capturing the actual economic opportunities available to the reporting entity. Most obviously, its emphasis on selling price rather than net realizable value (selling price less cost to sell) fails to measure the item in terms of the actual cash flow stemming from it. The use of exit (selling) rather than entry prices precludes the possibility that entry (purchase) prices may be more relevant in some circumstances. For example, replacement costs may better capture the cost of using assets, and therefore lead to more useful measures of profit margins, in a going concern business, in which replacement is the norm. Another potentially useful measure precluded by fair value is value-in-use, an entity-specific measure of the present value of the cash flows realizable by continued holding of the item.11 Mary Barth explains in a footnote that value-in-use, as defined in IAS 36, is not acceptable as fair value. This measure can be informative where it represents the most profitable use of the item from the entity's perspective, for example, where financial markets are illiquid so that the most profitable use of an asset such as a loan portfolio is to hold it to maturity. Of course, if markets were really perfect and complete, the question of illiquidity would not arise, and neither would the difference between entry and exit prices. If markets were also frictionless (no transaction costs) the issue of choosing price rather than value would also be irrelevant. An informational approach to measurement, as advocated above, has two broad implications. First, where markets are imperfect, accounting data aid decision making by users but are not alone sufficient to determine users' decisions. They provide inputs to users' models but do not provide complete information. This proposition is consistent with the fact that investors, who are important users of accounts, will have regard to a wide range of economic data and other non-accounting information in valuing their investments. It would be an unusual situation in which an appropriate valuation of an entire entity could be obtained simply by observing the fair value of the underlying assets and liabilities (Penman, 2007). Second, just as ‘different costs for different purposes’ has long been accepted as being appropriate for assisting management decisions (Clark, 1923), so ‘different measures for different purposes’ is appropriate for financial accounting. For example, as Penman (2007) demonstrates, cost measures may provide useful margins on turnover for predicting operating cash flows in a going concern business, whereas fair values may be a more direct and reliable means of valuing a portfolio of marketable investments. Many alternative measures might be potentially useful as information to some user, but it would be impractical, expensive and possibly confusing to report many alternative measures. For this reason, multiple-column reporting, advocated by Stamp (1979) and others, has never found favour in practice. There is therefore a need to provide some principle that will limit the range of the measures required under an informational approach. A first step towards this would to be to define a measurement objective more broadly than is often the case by allowing it to encompass different measurement methods in different circumstances. A method here is assumed to be a ‘pure’ measure embracing only one approach, such as historical cost, fair value or replacement cost. Thus, although only one method of measurement would be associated with a particular item, different items would, or could, be measured using different methods, if those methods best represented the economic properties of the particular item. In the context of the current FASB and IASB conceptual frameworks, such an objective should be designed to provide the information that is most relevant to assessing the current economic position and financial performance of the entity. Deprival value is a measurement objective that falls within this broad requirement, offering a more precise definition of how the objective should be achieved. It attempts to measure the economic opportunities facing the entity, by reporting the result of the opportunity that a profit-maximizing firm would select, in the specific circumstances facing it. Deprival value is therefore proposed here as a possible measurement objective, overcoming some of the possible objections to fair value and other ‘ideal’ measurement approaches. Richard Macve discusses deprival value in more depth. Here, we are concerned more with advocating an informational approach rather than a pure measurement approach, on the ground that a realistic market setting implies the former rather than the latter. Deprival value is consistent with our preferred approach but is not necessarily the best solution, although it is certainly the best developed in the existing literature. It should be acknowledged in particular that deprival value has its own problems of implementation. It requires three different measures for each item, including the highly subjective value-in-use. It also has some difficult aggregation problems (Edey, 1974). Moreover, the essence of the informational approach is that the appropriate measurement is dependent on the use, the circumstances (including market opportunities that are actually available) and the model of the user, so that no particular measurement objective can be expected realistically to meet all needs in all circumstances. Hence, the choice of measurement objective in this world of imperfect markets and asymmetric information is bound to involve a degree of subjective judgment. Therefore, we should be modest in our claims for specific solutions and realistic about the inevitable limitations of financial reporting. Richard Macve and Andrew Lennard have attached to their papers appropriate comments on the IASB Exposure Draft, Fair Value Measurement, which appeared after the symposium (28 May 2009). As they report, this repeats the deficiencies of the earlier Discussion Paper and, in some respects, reinforces them. In the context of the arguments presented earlier, the following three key points are worthy of particular note: The Exposure Draft retains the assumption that fair value is an exit value. It makes this more confusing by the reference to the use of replacement cost, justified by ‘the economic principle of substitution’, for tangible fixed assets (BC, para. 63). This sounds very much like deprival value, but BC, paras 65–6 attempt to dispel this view by stating that the Exposure Draft's proposals, unlike deprival value, are based on the non-entity-specific views of ‘market participants’. Unfortunately, in the absence of perfect markets from which to observe exit values, the concept of a ‘market participant’ requires the assumption that such individuals are fully informed and endowed with identical resources to those of the reporting entity, that is, they are effectively identical to the entity. This makes nonsense of the idea that the resulting measure is non-entity specific. In particular, by assuming that the market participant is fully informed, it ignores the importance of information asymmetry as the motivation for financial reporting: the whole point is that market participants are not fully informed. Fair value is still defined, in the Exposure Draft, as a price rather than a value that can be realized, because transaction costs are excluded. The reason given for their exclusion (BC, para. 47) is that transaction costs, unlike transport costs which would be included, are characteristics of the transaction rather than the asset. This is apparently inconsistent with the fact that fair value is defined as the price resulting from a sale transaction, and is therefore also a characteristic of the transaction. The contradiction is also apparent in the Exposure Draft's acceptance of transaction costs being deducted from selling price when identifying the ‘most advantageous market’ (8), although once the market is selected, fair value is the price unadjusted for selling costs.